Upload
others
View
25
Download
0
Embed Size (px)
Citation preview
Introduction
Words and phrases
Grammatical functions and case
marking
Morphology and syntactic valency
Periphrasis and constructional idioms
Conclusion
In dealing with the morphology-syntax
interface, there are four main thematic
areas to look at in order to understand how
these two linguistic domains interface.
We shall observe;
1.The demarcation of the empirical domain
of morphology and syntax:
•When is a multimorphemic sequence a word,
and when is it a phrase?
2.Morphology and syntax interact in two
ways:
•Syntactic constructs may form parts of a
complex word, and
•Syntax in its turn governs the use of
morphological case marking on words.
3.How morphological operations may affect
the syntactic valency of words.
4.Languages may have syntactic alternatives
to the morphological expression of
grammatical and semantic content, and we
might therefore want to know more about
the division of labour between morphology
and syntax in this respect.
Morphology deals primarily with the
structure of words, and syntax with the
structure of phrases. But how do you know
if a particular combination of morphemes is
a word or a phrase?
Is the lexical unit hard disk a word (that
is, compound of the type A + N), or a noun
phrase? How can we know?
The most important criterion for wordhood
is Lexical Integrity.
Lexical Integrity:
‘The syntax neither manipulates nor has
access to the internal form of words’
(Anderson 1992: 84).
Let’s take the English compound teapot
•It is impossible to take out one component
and topicalize it:
•*Tea, I bought pots.
• Teapots, I bought.
•Anaphoric devices cannot refer to parts of
the compound:
•*He took the teaipot and poured iti into the
cup.
A) Lexical Integrity implies that if we call
something a word, syntactic rules cannot
apply to its integral parts.
The importance of this criterion can be
illustrated by looking at the difference
between prefixed verbs and particle
verbs in Dutch.
Dutch has a syntactic rule of Verb
Second which requires finite verbal
forms in main clauses to appear in
second position, after the first
constituent.
However, the underlying word order in
Dutch is SOV (Subject–Object–Verb).
This means that in the surface form, the
verb moves.
If the verb is prefixed, the prefix moves
along with the stem, which shows that the
prefix is not a separate word.
On the other hand, if the verb is a
particle verb (that is a phrasal verb), when
the main verb moves, the particle is
stranded in the original position of the V.
This shows that particles are words on their
own.
Hence, we get the following contrast for the
Dutch
• prefixed verb doorzóeken “to search”
• particle verb dóorzoeken “to continue
searching”
a. Jan door-zocht het hele gebouw
John through-seek.PAST.SG the whole
building
“John searched the whole building”
b. Jan zocht tot 2 uur door
John seek.PAST.SG till 2 o’clock through
“John continued searching till 2 o’clock”
The difference between prefixed verbs
and particle verbs is reflected by the
difference in the location of the main
stress of these expressions:
doorzóeken & dóorzoeken
Particles bear main stress, whereas in
prefixed verbs it is the verbal stem that
carries main stress.
B) Lexical integrity also implies that
English verb particle constructions such as
to look up are to be considered phrasal
verbs because the two parts can be
separated:
John looked up the information.
John looked the information up.
Essentially, since syntax rules can break
it up, it violates lexical integrity and is
thus a phrase.
C) Lexical Integrity manifests itself in the fact
that syntactically governed rules of inflection do
not apply to the individual parts of a word.
Dutch:
•Ròd-eA ko’olN “red cabbage”
•This expression is a phrase because;
1. the adjective is inflected –e, as is the rule in
Dutch for prenominal adjectives.
2. the main stress is on the second constituent
kóol
•Zu’urAkòolN “lit. sour cabbage, sauerkraut”
•This expression is a compound word because;
1. The adjective is not inflected
2. the first constituent bears the main stress, as is the rule for Dutch nominal compounds
In English, a compound word is distinguished from a
phrase using stress
•Compound word: main stress is on the first
constituent. e.g. ‘greenhouse
•Phrase: main stress is on the second constituent.
e.g. green ‘house
D) The extent to which Rules of Anaphora
are subject to the Lexical Integrity constraint
is a debate.
When it comes to Anaphora, Lexical
Integrity constraint is seemingly limited.
Let’s take the following examples:
a) John likes [the guitar]i because he thinks iti
is a social instrument.
b) John became a guitarist because he
thought it was a social instrument.
In (a), [guitar] is co-indexed with ‘it’,
meaning they refer to the same entity. The
pronoun ‘it’ is interpreted as ‘guitar’.
In (b), the pronoun ‘it’ is also interpreted
as ‘guitar’.
However, we see ‘guitarist’ and not ‘guitar’ in (b). This would imply a co-indexation of the nature:
•John became a [guitar]i-ist because he thought iti was a social instrument.
If so, the Lexical Integrity constraint does not hold for rules of anaphora.
This however goes against the general observation that words, when embedded in complex words, lose their referential potential.
Rather, sentence (b) can be accounted for
like this:
• The pronoun ‘it’ receives an interpretation
within the domain of discourse evoked by
this sentence.
•The word ‘guitarist’ evokes the entity
‘guitar’, which is an entity in the discourse
domain, and thus becomes the referent or
interpretation of ‘it’
To demarcate between morphology and
syntax, there is the need for a formal
criteria. This is not exactly straight forward.
Take for instance the expression ‘yellow
fever’
Semantically, it is a lexical unit and is
stored in the English lexicon as such.
Morphologically, it is not really a word. Its
stress pattern is that of a phrase with the
main stress on the head word: yellow
‘fever.
Sometimes there is no clear demarcation
between between syntax and morphology.
Is there a NO PHRASE CONSTRAINT on
complex words?
•Answer: NO
•Phrases do occur as part of complex words.
Look at these examples of complex words from English:
•[special exhibitions] gallery
•[module for module] upgrade
•[drugs and rehabilitation] centre
The words in square brackets comprise phrases.
In the first example, [special exhibitions] is a phrase of the type A+N with the main stress on the second constituent.
However, not every phrase can contribute
to word formation.
For example, English phrases that have
determiners cannot be part of complex
words.
• *the [[the special exhibitions] gallery]
•This ungrammaticality has to do with the
fact that [the special exhibitions] with the
definite determiner is a referring expression,
whereas word constituents in non-head
position have a modifying, classificatory
function.
Affixation of phrases is possible, but it is
limited to highly productive affixes such as:
•The English nominalizing suffix –er
-E.g. do good-er, fast track-er, do-it-
yourself-er
•The Dutch equivalent –er
-E.g. vierde klass-er “fourth grader”
•The Dutch diminuitive suffix –(t)je
-E.g. twaalfuur-tje
“lit. twelve o’clock-DIM; lunch food”
From the above discussion we see how
morphology is demarcated from syntax.
However:
•Syntax can feed word formation.
•Morphology also feeds syntax by supplying the
units to be operated on by syntactic rules.
GRAMMATICAL FUNCTIONS
AND CASE MARKINGS
In most languages, the interface between the
semantic properties of a clause and its morpho-
syntactic structure (word order, case marking) is
partially regulated by the grammatical function
frame of the verb of that clause.
For instance, the English verb HIT denotes an
action with two participants: the one who hits,
and what/who is hit.
We refer to these crucially involved entities as
core arguments of the predicate ‘hit’
The Predicate Argument Structure (PAS) of this
verb can be represented as Hit x,y
•E.g. Kofi hit Ama.
PAS refers to the number and type of
participants/ arguments a predicate requires.
This relationship between the PAS of a verb
and its syntactic realization can be expressed
by linking the grammatical functions
‘subject’ (SUBJ) and ‘object’ (OBJ) to the
arguments of PAS.
HIT, xAGENT, yPATIENT Predicate Argument Structure
SUBJ OBJ Grammatical Function Frame
The two arguments of to hit are the core
arguments that always have to be expressed.
The number of arguments a verb requires to be
expressed is referred to as its syntactic valency.
In addition to the core arguments, there might be
other entities involved in specifying the event of
‘hitting’ such as the instrument and the location.
•e.g. John hit his enemy in the back with a
stick.
These latter specifications are always
optional, and such participants in the event
are usually called adjuncts.
The relationship between the two levels of
PAS and grammatical functions is often
predictable by linking rules.
If there are two arguments, then the argument
that expresses the Agent of the action will be
expressed as the grammatical subject, and the
other argument as the grammatical object.
If there is only one argument, the default linking
rule applies:
•When there is an intransitive verb, and the lone
argument will be linked to the grammatical
function of subject.
Linguists classify languages in terms of the order in which grammatical functions are expressed:
•SVO (Subject–Verb–Object), SOV (Subject–Object–Verb), etc. That is, we need these grammatical functions for syntactic purposes.
In many languages, morphology is used to mark grammatical functions, either through head marking or through dependent marking.
Case marking is a form of dependent marking
which signals the grammatical function of an
NP in a clause.
Indo-European languages with morphological
case systems show the distinction between
grammatical subject and grammatical object
by means of the opposition between the
nominative case and the accusative case.
If there is only one argument, this
argument is case-marked as a nominative.
When there are two arguments, the
subject is marked as nominative and the
object as accusative.
This system is called the Nominative-
Accusative system.
An alternative case system is the
Absolutive-Ergative system, used in many
Australian languages.
The following symbols are used (Dixon 1994: 6-
7):
S= Intransitive subject
A= Transitive subject
O= Transitive object
In the Nominative- Accusative system A and S
receive the same case marking.
In the Absolutive- Ergative system O and S
receive the same case marking:
•This system marks the object of transitives and the
subject of intransitives using the same case:
absolutive; subjects of transitives are marked
with a different case: ergative.
• PAS PREDICATE, x PREDICATE, x y
Grammatical
function
S A O
Nominative-
Accusative
system
NOM NOM ACC
Absolutive-
Ergative
system
ABS ERG ABS
Dutch
•Der Mann lach-t
The NOM man NOM laugh-PRES.3SG
“the man(S) laughs”
•Der Mann kauf-t ein Buch
The NOM man NOM buy a ACC book ACC
“the man (A) buys a book (O)”
Dyirbal
• ŋuma nanaga-nyu
Father ABS return- NONFUT
“Father (S) returned”
•yabu ŋuma-ŋgu bura-n
Mother ABS father ERG see- NONFUT
“Father (A) saw mother (O)”
In some cases, it is not only the case marking that
is ergative in nature, but also the syntax; as can be
seen in the construction of elliptical sentences.
In ellipsis, the second of two identical NPs in
coordinated clauses is omitted, as illustrated below
for English:
•E.g. John (A) saw his wife (O) and—(S) rejoiced.
•Here the transitive subject (A) and the intransitive
subject (S)- which has been omitted- refer to the
same NP.
Essentially, we see that the case marking that
an NP takes in a clause determines the
grammatical function that it occupies on the
syntactic level and vice versa:
•Syntactic constructs may form parts of complex
words (morphology), and syntax in its turn
governs the use of morphological case marking
on words.
MORPHOLOGY AND
SYNTACTIC VALENCY
The relation between the PAS and the level
of grammatical functions can be changed by
morphological operations.
The best known and widely studied example
of such a change is PASSIVIZATION.
In this operation, the agent of the
predicate is demoted to the status of an
adjunct.
The remaining argument, if any, will then
receive the status of S, and receive the nominative case in the nominative-accusative system.
This is illustrated in Greenlandic Eskimo:
• a) inuit nanuq people.ERG.PL polar.bear.ABS.SG
taku-aat
see-3PL.3SG.IND
“The people saw the polar bear”
b) nanuq (inun-nit)
polar.bear.ABS.SG people-ABL.PL
taku-niquar-puq
see-PASS-3SG.IND
“The polar bear was seen (by the people)”
In (b), the passive suffix has the effect of
making the original subject optional.
It can be added as an adjunct, marked with ablative case; whereas in the active sentence it is marked with the ergative case.
In (a), the person, number, properties of both the plural agent and the singular patient are marked on the verb.
In (b) it is only the properties of the singular patient that are marked on the verbal form.
• This is because of the non- argument status
of the Agent in (b).
In present-day Romance and Germanic
languages like French and English, there is
no synthetic passive form of verbs.
Instead, a periphrastic construction is
used, consisting of a passive auxiliary and a
participle.
E.g.
•Je suis insult-é par Jean
I be 1SG.PRES insult-PAST.PTCP.MASC.SG by Jean
“I am insulted by Jean”
Passivization is an operation that affects the
mapping between PAS level and the grammatical
function level.
It is not an operation on the semantic level since it
does not change meaning.
Only the form in which the meaning is
expressed is changed.
This is also proven by the fact that the
demoted Agent of the verbal predicate is
semantically still available, for instance as
a “controller”.
•Example:
a) The boat was sunk to collect the insurance money.
b) *The boat sank to collect the insurance money.
With the passive sentence in (a), the
demoted agent is still there semantically, as
the implicit subject of the verb [collect] in
the embedded infinitival clause.
In (b), the intransitive, non-passive verb
[sink] is used, and hence there is no agent
involved at all.
Consequently, an appropriate controller of
the subject of [collect] is not available in
sentence (b), making it ungrammatical.
The essence of passivization is the
demotion of the Agent argument, not the
promotion of the Patient argument.
The promotion of the Patient to the status
of grammatical subject is as a result of the
default linking rule.
The default linking rule requires that if
there is only one argument, it should be
expressed as the grammatical subject.
Dutch and German exhibit passivization of
intransitive verbs. (Impersonal passive)
E.g. Er werd enthousiast
there AUX.PASS.PAST.IMPF enthusiastically
gedanst.
dance PAST.PTCP
“There was enthusiastic dancing.”
In the above sentence, the agent is no longer
mentioned explicitly, and since the verb dansen
“to dance” is intransitive, there is no other
argument that can be linked to the subject
position.
However, since Dutch clauses with finite verbs
always require the presence of a subject, the
subject position is filled with a dummy word: er.
The different patterns of linking arguments
to grammatical functions are also referred
to as different VOICES:
•Active voice
•Passive voice
•Middle voice
Voices express the semantic relations
between the subject and the action
described by the verb.
The grammarians of Sanskrit and Greek speak
of the middle voice as a distinct verbal form,
just as we may have different aspectual forms of
a verb.
The use of the middle voice instead of the
active voice does not mean that the mapping of
semantic roles on grammatical functions is
different.
However, the middle voice marker changes the status of the subject with respect to the denoted action. Below is an example from Sanskrit:
a) Devadattah katam karoti
Devadatta.NOM mat.ACC make.SG
“Devadatta makes a mat”
b) Devadattah katam krute
Devadatta-NOM mat.ACC make.SG
“Devadatta makes himself a mat”
Thus, as the example shows, the middle
voice indicates that the subject is also the
beneficiary of the action expressed by the
verb.
ANTI-PASSIVE CONSTRUCTION
Instead of the Agent, the O (transitive object) can
also be demoted. This is the anti-passive
construction, found in particular in languages with
an absolutive-ergative system.
In this construction, the O can be omitted, or
appear with instrumental case. Hence, you get an
intransitive sentence.
The following example is from Greenlandic
Eskimo:
a) inuit tuqup-pai
people.ABS.PL kill-3SG.3PL.INDIC
“He killed the people”
b) inun-nik tuqut-si-vuq
people.INSTR kill-ANTIPASS-3SG.INDIC
“He killed people”
• Sentence (a) exhibits the normal case marking.
• In (b) we have an intransitive sentence, with
the word for “people” marked with an
instrumental case.
The effect of the anti-passive is to
‘despecify’ the direct object of the
transitive verb, if it is expressed at all.
Hence, (b) does not refer to specific
people that were killed.
Note also that the person–number
properties of “people” are no longer marked
on the verb. It is only the instrumental case
that is marked.
APPLICATIVE CONSTRUCTION
Instead of demoting an argument, you may also promote it.
This is the applicative construction that is found in many African and Austronesian languages.
For instance, the effect of adding an applicative affix to the verb may promote an instrumental or locative NP to the status of O(transitive object).
E.g. Atlantic language, Wolof.
a. Mungi lekk ag kuddu
PRES.3SG eat with spoon
“He is eating with a spoon”
b. Mungi lekk-e kuddu
PRES.3SG eat-APPL spoon
“He is eating with a spoon”
Also, an applicative affix can introduce an
additional beneficiary argument.
E.g. Mexican language, Classical Nahuatl:
• Ni-c-no-pa¯qui-lia
I-it-myself-wash-APPL
“I wash it for myself”
NOUN INCORPORATION is another way of
making transitive verbs intransitives.
It is when an argument of the verb is not
expressed by a separate NP, but as part of a
verbal compound. The verbal compound
then functions as an intransitive verb.
The following example, a one-word-
sentence, comes from the Amerindian
language Tuscarora:
• w-e-khw-əti-ʔ
FACTUAL-FEM.AGENT-food-make-PERF
“she meal-made = she cooked”
In Tuscarora, verbs with incorporated nouns co-
occur with verbs with independent noun phrases.
The difference is that the verb with incorporated
noun denotes an institutionalized action, whereas
the independent noun phrases have referential
potential.
This functional difference is nicely illustrated by
the fact that in Tuscarora you can have
constructions of the type She bread-made corn,
with both an incorporated noun and an independent
NP, meaning “She made corn-bread”
(Mithun 1999: 46).
The nature of Noun Incorporation;
Suppose we had an English verbal compound:
wood-chop
[Chop] is the head of the compound and
[wood] is the incorporated NP.
The PAS of chop is Chop x,y
The NP y is [wood] which is incorporated
into the verbal compound.
The only other NP, x takes up the position of
the grammatical subject, and thus, the verb
behaves as an intransitive.
What we see here, is that the semantic
interpretation of a complex verb may lead to a
syntactic valency that is different from that of
its verbal head.
The phenomenon of noun incorporation has led
some linguists to propose that some kinds of
word-formation can be accounted for by
syntactic operations.
Here, the incorporated noun is represented
as an independent NP at the underlying
syntactic level.
The noun is then moved to be adjoined to
the V in the surface representation,
resulting in a verbal compound.
For instance, the fictive English verbal
compound “to bed-buy” would be derived
as follows:
S S
NP VP NP VP
V NP V NP
N Ni V ti
I buy bed I bed-buy
The noun bed originates in the object
position of the VP, which is then vacated.
The empty position is indicated by t (for
trace), and is co-indexed with the moved N.
Through this co-indexation, the
incorporated N will be interpreted
semantically as the object of the verb.
And so in this analysis the intransitivization
effect of this kind of noun incorporation is
the effect of a syntactic operation.
-Baker 1988, 1996, for the Amerindian
language Mohawk.
CAUSATIVES
The choice of a morphological or a syntactic
analysis of word-formation plays a role in the
analysis of causatives.
Causative verbs are verbs where the A has
the role of the causer of an event in which
one or two entities play a role.
A classic example of a simplex causative verb
in English is to kill, with the meaning “cause
to die”.
The semantic structure of this predicate can be represented as follows:
CAUSE (x, (DIE, y))
Sentences with causative verbs denote complex events.
Many languages have causative affixes that turn
non-causative verbs into causative ones. The
semantic effect is the addition of a predicate CAUSE
and an additional argument, the ‘CAUSER’.
Hence, causativization has the effect of
increasing the valency of words. This applies
to adjectives, nouns and verbs.
E.g. Diyari, an Australian language:
•kidi “clever”(adj.) > kidi-ŋanka “to teach =
to make clever”
•muka “sleep”(noun) > muka-ŋanka “to put to
sleep”
Examples from Turkish
•Müdür mektub-u imzala-dt ˙
director letter-DO sign-past
“The director signed a letter”
•disçi mektub-u müdür-e imzala-t-tt ˙
dentist letter-DO director-IO sign-CAUS-PAST
“The dentist made the director sign a letter”
This example shows how the addition of a
causative suffix increases the valency of a verb. It
is an example of a valency- increasing operation.
Note that, causativization is a morphological
process that primarily affects the semantic
properties of a predicate, and hence the level of
PAS.
A CAUSER argument is added to the PAS of the
input word, which affects its syntactic valency
through the linking rules that map arguments onto
the grammatical function frame of a verb.
A syntactic approach to causativization assumes
that the causative suffix functions as a verb with a
complement that denotes the caused event.
However, the causative verb is a bound morpheme
so it cannot surface as a word of its own, and must
be combined with a verbal stem.
The verb of the embedded clause is moved to the
higher clause, and attached through adjunction to
the left of the causative suffix that functions as the
verb of the main clause.
S S
NP VP NP VP
S V S V
NP V NP V V V
Ali Hasan öl dür Ali Hasan ti öli –dür
Ali Hasan die CAUS “Ali killed Hasan”
In sum, verbs may carry morphological
markings that determine how their
arguments must be expressed on the level
of syntactic structure. The syntactic valency
of verbs may thus be affected by
morphology.
Periphrasis is used in situations where
single morphological forms of certain
constructions cannot be found. Instead, a
word combination has to be used:an analytic
form.
Constructional idioms are multi-word
expressions that are idiomatic in nature but
not completely fixed because some positions
are variable.
Examples of periphrasis English are the use
of auxiliaries + participles of main verbs
to express the perfect tense like; “had
called” and the passive voice; “was
called”.
In a number of languages, the progressive
aspect is expressed by a periphrastic form
of the verb ‘to be’ + prepositional phrase,
as illustrated by the following examples
from Dutch:
• Jan is [[aan]P [[het]Det [fiets-en]N]NP]PP
John is at the cycle-INF
“John is cycling”
• Jan is de aardappels aan het schillen
John is the potatoes at the peel-INF
“John is peeling the potatoes”
It is key to note that in the second
example, the direct object de aardappels
“the potatoes” is not located right before
the verb, as is normally the case for objects
in Dutch embedded clauses.
Instead, the object precedes the word
sequence aan het “at the” that signals the
progressive aspect.
Progressive aspect may also be expressed
by using postural verbs such as “to sit”
and “to stand” in coordination with a
main verb, as illustrated here for Afrikaans
and for West-Flemish, a dialect of Dutch:
Afrikaans
Piet staan ’n glas water en drink
Pete stands a glass water and drink
“Pete is drinking a glass of water”
West Flemish
Zij zat kousen en stoppen
She sat stockings and mend-INF
“She was mending stockings”
Note that it is only the postural verb in the
West Flemish sentence that has a finite form,
the main verb appears in the infinitive. (In
Afrikaans there is no formal difference
between infinitive and finite forms).
Such constructions with a periphrastic function
are constructional idioms.
Many languages have PREVERB + VERB
combinations that function similarly to prefixed
verbs.
The notion preverb refers to words that appear
before verbs, and form a close unit with that verb.
Quite often, these phrasal constructions function
as alternatives to prefixed verbs.
A Dutch example of a preverb+verb construction is
this: af V with the meaning “to finish V-ing”.
af is a Dutch particle that expresses a result. It
turns (in the following example) the
intransitive verb werken “to work” into a
resultative predicate:
• Bettelou werkte haar opdrachten af
Bettelou worked her assignments PARTICLE
“Bettelou finished her assignments”
Particle verbs are lexical units, but not words
in the morphological sense.
Analytic causative constructions are also
instantiations of constructional idioms.
In Germanic and Romance languages, the
causative meaning is often not expressed by an
affix, but by a separate causative verb such as
laten “to let” in Dutch and fare “to do” in
Italian.
The combination of the causative and the main
verb functions as a unit, and the recipient is
marked by a preposition, aan in Dutch, and a in
Italian:
Dutch
Ik liet het boek aan mijn collega zien
I let the book to my colleague see-INF
“I showed the book to my colleague”
Italian
Ho fatto vedere il libro a-l mia have.1SG make-PTCP see.INF the book to-DEF my
collega
colleague
“I showed the book to my colleague”
The unitary nature of liet zien and ho fatto
vedere manifests itself in the fact that they
select a recipient argument marked by a
preposition, whereas neither the causative
verb nor the main verb select a recipient on
their own.
Another type of constructional idiom that is
functionally similar to complex verbs is that
of SERIAL VERBS, found in many African,
Austronesian, and Papua languages.
Characteristics of such constructions are
that the two (or more) verbs denote a single
event. There is only one overt subject, and
one tense marker.
São- Tomense (Portuguese- based creole)
• Bisu vwa subli
bird fly.PAST go.up
“The bird flew upwards”
• Zon toma mantchin kota po
Zon take.PAST machete cut tree
“Zon cut the tree with the machete
In sum, languages may have syntactic,
analytic alternatives to the morphological
expression of meaning. These syntactic
alternatives may have the status of lexical
units and may exhibit special syntactic
behaviour.
CONCLUSION
We see from the above discussion that
Morphology does not just interact with Syntax,
these two branches of linguistics actually
interface. That is, they influence one another.
Morphology can dictate how Syntax is
represented or expressed, and vice versa.